Cross-Context Event Contracts: Backoffice → Commerce, Operations

Formal event contracts for events emitted by the Backoffice bounded context. For the pricing-specific events (PriceMatrixPublished, BookingConfirmed), see event-contracts-pricing.md. For the broader domain events catalog, see domain-driven-design.md §4.

---

Event Catalog

Event	Emitter	Consumer(s)	Purpose
TripPublished	Backoffice (TourDeparture.status → PUBLISHED)	Commerce (TourOffering creation, boarding points + ancillaries sync), Operations (ServiceLeg creation)	Signals a departure is ready for sale and dispatch
TripCancelled	Backoffice (TourDeparture.status → CANCELLED)	Commerce (TourOffering.status → CANCELLED)	Stops sales immediately

---

Payload Schemas

TripPublished

The most cross-cutting Backoffice event. Each consumer reads the fields it needs; the payload carries the full departure context to avoid follow-up queries.

Field	Type	Description
event_id	UUID	Idempotency key
tenant_id	UUID	Owning operator
tour_departure_id	UUID	The published departure
tour_template_id	UUID	Source template
start_date	DATE	Departure date
end_date	DATE	Return date
capacity	INT	Max passengers (from capacity_rules)
max_door_pickups	INT	Door pickup limit (from tour_templates.max_door_pickups)
deposit_config	JSONB	Resolved deposit config (template override or tenant default)
cancellation_policy	JSONB	Resolved cancellation policy
boarding_points	TripPublishedBoardingPoint[]	Resolved boarding point catalog (library defaults + template overrides + is_origin surcharge zeroing). See boarding-points.md §Commerce Projection.
ancillaries	TripPublishedAncillary[]	Resolved ancillary catalog (catalog defaults + template overrides). See ancillary-catalog.md §Commerce Projection.
legs	TripLegDefinition[]	Itinerary structure for ServiceLeg creation. See schema-operations.md §service_legs for the Operations consumer handler.
published_at	TIMESTAMPTZ	Publication timestamp

TripPublishedBoardingPoint fields: boarding_point_id, name, address, geo_coordinates, zone_label, surcharge, is_origin, door_pickup_available, door_pickup_surcharge, door_pickup_radius_km, passenger_instructions, display_order.

TripPublishedAncillary fields: catalog_item_id, type, label, description, cover_image_key, price, currency, is_per_passenger, max_quantity, included_by_default, sort_order.

TripLegDefinition fields: sequence_order, leg_type, scheduled_start, scheduled_end, waypoints[] (each: geo_coordinates, sequence_order, label, waypoint_type).

Consumer handler logic:

• Commerce: Creates/updates tour_offerings row. Denormalizes boarding_points → available_boarding_points JSONB, ancillaries → available_ancillaries JSONB, max_door_pickups, deposit_config, cancellation_policy. Idempotent on re-publish (upsert on tour_departure_id).
• Operations: Creates service_legs + route_waypoints from legs[]. Uses UPSERT ON CONFLICT (tour_departure_id, sequence_order). Skips legs in ACTIVE/COMPLETED/CANCELLED status. See schema-operations.md §service_legs for the full handler contract.

TripCancelled

Field	Type	Description
event_id	UUID	Idempotency key
tenant_id	UUID	Owning operator
tour_departure_id	UUID	The cancelled departure
tour_template_id	UUID	Source template
cancelled_at	TIMESTAMPTZ	Cancellation timestamp

Consumer handler logic:

• Commerce: Sets tour_offerings.status → CANCELLED for the matching tour_departure_id. The booking widget immediately stops selling this departure. Existing bookings are unaffected (operator contacts passengers manually at V0.1; automated notifications via Communications arrive at [v0.2]).

---

Delivery Contracts

Event	Mechanism	Trigger	Idempotency
TripPublished	Hasura Event Trigger (async, post-commit)	backoffice.tour_departures.status → PUBLISHED	event_id (UUID). Commerce: upsert on tour_departure_id. Operations: upsert on (tour_departure_id, sequence_order).
TripCancelled	Hasura Event Trigger (async, post-commit)	backoffice.tour_departures.status → CANCELLED	event_id (UUID). Commerce: idempotent status set (already CANCELLED = no-op).

All consumers implement at-least-once processing with event_id-based deduplication.